I think some of the force of the presentation was lost, because C++ was the whipping boy of choice. It's a pretty low-level language, and finding problems with it is a bit too easy. A better idea, IMO, would be to look at large bodies of high-level-language code (like Smalltalk or Perl) and see what repetitive patterns emerge in those languages. After all, those languages already fixed the obvious problems in C and co. -- the question is what are the new bottlenecks?